Re: Soot 2.0.0 info request

[Ondrej Lhotak <olhotak@sable.mcgill.ca>]

On Fri, May 09, 2003 at 09:34:09AM +1200, Stephen Cheng wrote:
> Cool Ondrej, I am looking forward to it.
> 
> >- a true whole-program mode
> How is this different from current whole program mode? 

Well, there isn't really a current whole program mode. There's
single-file mode, and there's application mode, which just does
intraprocedural analyses on the methods reachable from a given starting
method. There isn't currently a mode that guarantees that you will have
a call graph available, and that Soot will have some idea of all the
methods that are reachable. So, this is what we're adding: a -w switch
that switches to whole-program mode, and certain phases that are enabled
only in this mode, and which are guaranteed a call graph (which can be
created using CHA, or Spark, or whatever).

> >- first release of Eclipse integration (many sub-features in here)
> Excellent. I have AspectJ with Eclipse as well. I would look forward to use
> Dava with AspectJ output.

You mean extending Dava to output aspects? That sounds interesting...

> >- SSA-Jimple. Really, this time.
> Have the optimisations converted to use Jimple-SSA?

I don't think we planned to convert the existing optimizations to SSA.
We mainly just wanted to provide SSA to allow people who want to write
new optimizations to use SSA if they want (and to optionally use it
as an input to Spark). But, the SSA is very similar to Jimple, so it
shouldn't be too hard to convert the existing optimizations to it.
I think Navin has implemented at least some basic SSA optimizations,
so those will be included.

Ondrej